1

Battery Energy Storage System Models forMicrogrid Stability Analysis and Dynamic

SimulationMostafa Farrokhabadi, Student Member, IEEE, Sebastian Konig, Claudio Canizares, Fellow, IEEE,

Kankar Bhattacharya, Fellow, IEEE, and Thomas Leibfried, Member, IEEE

Abstract—With the increasing importance of battery energystorage systems (BESS) in microgrids, accurate modeling plays akey role in understanding their behaviour. This paper investigatesand compares the performance of BESS models with differentdepths of detail. Specifically, several models are examined: anaverage model represented by voltage sources; an ideal dc sourcebehind a voltage source converter; a back-to-back buck/boostand bi-directional three phase converter, with all models sharingthe same control system and parameters; and two additionalproposed models where the switches are replaced by depen-dent sources to help analyze the differences observed in theperformance of the models. All these models are developed inPSCAD and their performances are simulated and comparedconsidering various issues such as voltage and frequency stabilityand total harmonic distortion, in a benchmark test microgrid. Itis shown through simulation results and eigenvalue studies thatthe proposed models can exhibit different performance, especiallywhen the system is heavily loaded, highlighting the need for moreaccurate modeling under certain microgrid conditions.

Index Terms—Energy storage systems, dynamic simulation,microgrids, modeling, stability.

I. INTRODUCTION

M ICROGRIDS are defined as a cluster of interconnecteddistributed energy resources (DERs), energy storage

systems (ESS), and loads which can operate in parallel withthe grid or in an islanded mode [1]. Under the smart gridparadigm, microgrids are considered a critical link in the evo-lution from vertically integrated bulk power systems to smartdecentralized distribution networks [2], with high penetrationof renewables, easily scalable structures, and increased reli-ability levels [3]. However, such active distribution networkspresent various challenges, such as the integration of plug-in electric vehicles and the proliferation of renewable energysources (RES), which have to be addressed when significantpenetration levels are reached [4]. In this context, BESS can bepractically helpful to deal with such challenges, as explainednext.

This work has been partially supported by NSERC-Canada through Dis-covery Grants.

M. Farrokhabadi, C. A. Canizares, and K. Bhattacharya are with theDepartment of Electrical and Computer Engineering, University of Wa-terloo, Waterloo, ON N2L 3G1, Canada (e-mail: m5farrok@uwaterloo.ca;ccanizares@uwaterloo.ca; kankar@ece.uwaterloo.ca).

S. Koenig and T. Leibfried are with Institute of Electric Energy Systems andHigh-Voltage Technology (IEH), Karlsruhe Institute of Technology, Karlsruhe,Germany (e-mail: sebastian.koenig@kit.edu; thomas.leibfried@kit.edu).

For several interesting technical features, BESS have re-ceived considerable attention recently, particularly as a solu-tion to the challenges facing modern active distribution net-works and microgrids. BESS can provide several key ancillaryservices, such as load shifting, dynamic local voltage support,short-term frequency smoothing, grid contingency support,and reduce the need for fossil-fuel-based generation [4], [5].Therefore, BESS are considered a key enabling element ofmodern smart grids and microgrids.

Since many utilities and researchers use simulation softwarepackages to model and investigate various issues in microgrids,grid components need to be adequately modeled to properlyreflect the behaviour and performance of the system. Severalcomponents of microgrids have been extensively studied, and avariety of models have been developed and reported in the lit-erature. However, since inverter-based BESS are relatively newelements of microgrids, fewer studies have been conducted ontheir modeling and control. In [6], a generalized mathematicalmodel of ESS is presented for voltage and angle stabilityanalysis based on the balanced fundamental-frequency modelof the Voltage Source Converter (VSC) and the dynamics ofthe dc link, based on a set of linear differential algebraicequations. However, the impact of inverter switching on thesystem performance is not modeled, and the performance ofthe proposed model under different microgrid conditions suchas heavy and/or unbalanced loading are not studied. DetailedESS models for transient analysis in microgrids are presentedin [5] and [7]. However, the focus of these papers is on ESSapplications in microgrids, without considering the impact ofESS modeling on the system dynamic performance. Simplifiedmodels of ESS are presented in [8] and [9], but similarto the aforementioned papers, they dwell on applications ofESS in power systems rather than the modeling of ESS; theESS models consist of only active and reactive power loops,neglecting VSC and dc link circuits, and the ESS dynamicperformance.

Considering the variety of BESS components, such as dc-dcand dc-ac converters, studies on the impact of level of detail inBESS modeling on the stability and dynamic performance ofmicrogrids have not yet been reported in the literature. Hence,considering the importance of BESS in active distributionnetworks and microgrids, this paper investigates and comparesmicrogrid dynamic performance using BESS models with dif-ferent depth of detail. Specifically, several models are studied:an average model represented by voltage sources [10]; an ideal

IEEE TRANSACTIONS ON POWER SYSTEMS, ACCEPTED JULY 2017

2

dc source behind a voltage source converter [11]; and a back-to-back buck/boost and bi-directional three phase converter[12], with all models sharing the same control systems andparameters; and two additional proposed models where theswitches are replaced by dependent sources to help analyzingthe differences observed in the performance of the models.The models are developed in PSCAD and their performanceis compared considering various variables such as voltages andcurrents at the point-of-common-coupling, active and reactivepower injection, and total harmonic distortion, in the contextof the impact on the the stability of the system. Thus, the mainobjectives and contributions of the paper are the following:

• Develop and study various BESS models for microgridsimulation and analysis, including a new and efficientmodel where the switches are replaced by dependentsources, identifying the conditions in which each modelcan be used. A systematic mechanism is applied to isolateand evaluate the impact of various parameters.

• For the first time, study the impact of BESS con-verter switching on microgrid stability, including high-frequency signals and internal resistances of switches.

• Study the impact of BESS dc link voltage dynamicson microgrid stability, through eigenvalue studies anddynamic simulations.

• Study and demonstrate the impact of unbalanced loadingon microgrids stability.

• Compare the performance of the proposed models in arealistic test microgrid, determining their advantages andlimitations, and their computational efficiency.

The rest of this paper is organized as follows: Section II de-scribes the various components and parameters used to modelBESS in detail, and Section III presents three approximatedmodels to approximately represent BESS in dynamic systemstudies. Case studies and simulation results are presented inSection IV, comparing the performance of BESS models withdifferent depth of detail. Section V analyzes in detail thereasons for the differences observed in the performance ofthe various models presented. Finally, the main findings andconclusions of the studies presented in the paper are providedin Section VI.

II. BESS DETAILED MODEL

The BESS models used in this paper mainly include abuck/boost converter, a dc link capacitor, a three-phase bi-directional dc-ac converter, an ac filter, and a transformerconnecting the system to the microgrid. In this section, thesemodels and the corresponding parameters are discussed, andthe control techniques used for each BESS converter are alsodescribed. Figure 8 depicts in detail the BESS components [5].

A. Battery

The battery model described here is based on the genericmodel proposed in [13], and is modeled as a controllable idealdc source in series with an internal resistance RB . The no-loadvoltage of the battery EB is calculated based on the state-of-charge (SOC) of the battery using a nonlinear equation, asfollows:

EB = E0 −K1

SOC+ A−BQ(1−SOC) (1)

where E0 is the battery constant voltage in V, K is thepolarization voltage in V, Q is the battery capacity in Ah,and A and B are parameters determining the charge anddischarge characteristics of the battery. The parameters A,B,and K can be tuned to mimic a specific battery type dischargecharacteristic.

Note that the time frame of the test scenarios discussed inthis paper is in the order of a few seconds; hence, since theBESS designed and simulated in this paper has a rated powerof 1 MW and a rated capacity of 1 MWh, which is a typicalpower to capacity ratio for BESS in microgrids, the batteryvoltage and SOC relation cannot be observed in the presentedsimulation results.

B. Buck/Boost Converter

The buck/boost converter is in charge of controlling the dclink capacitor voltage by properly charging and dischargingthe battery. A cascaded PI controller is used here to generatethe duty cycle for the switches based on the difference of thedc link voltage and its set-point, as shown in Fig. 2. Note thatthe second PI controller reacts to changes in the active powerof the system first, thus improving the ability of the buck/boostcontroller to maintain the dc link voltage.

When the dc link voltage is lower than the pre-definedset-point, the converter works in the boost mode, dischargingthe battery. When the dc link voltage is higher than the set-point, the converter operates in the buck mode, charging thebattery. The current ripple is bounded by a proper choice ofthe inductor Lchopf , as follows [14]:

∆I =EB

Lchopf∆t =

EB

Lchopf

1

2fs(2)

where fs is the switching frequency.

C. DC-AC Converter

Figure 8 illustrates how the dc-ac converter connects thebattery and buck/boost converter to the grid through theac filter. The converter control system provides the voltagemagnitude and phase set-points to create sinusoidal referencesignals for the Pulse Width Modulation (PWM) scheme of theconverter.

As shown in Fig. 8, when the switches are in State 1, thevoltage magnitude and phase set-points are obtained basedon the reference voltage and frequency; thus, the BESS inthis case is the master voltage and frequency controller ofthe system, which is referred to as a grid-forming controltechnique. If the switches are in State 2, the voltage magnitudeand phase set-points are obtained based on the reference activeand reactive powers; hence, the battery is operated in constantPQ mode, injecting/absorbing constant active and reactivepower, which is referred to as grid-feeding mode. Grid-supporting and grid-following control modes are based onthese two fundamental control strategies, and are accomplishedby changing the reference voltage, frequency, active power,

3

Voc

Vga

Vgb

Vgc

Vdc

Ichop Idc

Lf Rfa

b

cRB

EB

Lchopf Rchopf

Buck/BoostB

atte

ry E

ner

gy

Sto

rag

e S

yst

em S

chem

e DC Link Inverter AC Filter Grid

PLL

abcdq0

Active and

Reactive Power

Calculation

Current

closed-loop

control

Current Reference

Generator

Ig_dq

Q

P

Qref

abcdq0

Vref_dq

Voa

Vob

Vref_abc

Modulation Technique

Vdc_meas

DC Voltage

closed-loop

control

DC Voltage

closed-loop

control

P

PrefPref

Modulation Technique

Mea

sure

men

t an

d C

on

tro

l S

chem

eBattery

Cf

Rd

V-F Reference

Generator

1

2Vref_dq

ref

2 1

refPLL

fref Vref

Vg_dq

PLL

aO

V

bO

V

cO

V

agI

bgI

cgI

cgVbgV

agV

dqgVdqgI

dqrefV

dqrefV

abcrefV

dcrefV

dcmeasV

1

2

3

4

5

6

7

8

n

d1

d2

N

Fig. 1. Schematic of a battery energy storage system components and its controls [5].

dcratedV

dcV

p iK K s+

-

drefVP

dc dcV I

' 'p iK K s+

-

+

Fig. 2. Buck/boost dc link voltage controller.

dratedV

dgV

p iK K s+

-drefV

qrefV0

1s refref

Fig. 3. Grid-forming voltage and phase reference generator.

and/or reactive power [15]. Observe that these controls arebased on Park’s dq-axes transformations.

In the grid-forming mode, the voltage dq-axes referenceset-points would be directly used to create the abc-referencesignals, as shown in Fig. 3. The reference angle is obtained byintegrating the reference angular frequency. Observe that a PIcontroller is used to maintain the point of common coupling(PCC) voltage at its rated value.

In grid-feeding control mode, the injected active and reac-tive power are calculated first, as follows:

p =3

2

(VgdIgd − VgqIgq

)(3)

q =3

2

(VgdIgq − VgqIgd

)(4)

To obtain the corresponding fundamental P and Q compo-nents, the instantaneous active p and reactive q powers arepassed through low-pass filters. The fundamental active andreactive powers are then passed through the current referencegenerator block to obtain the current dq-axes reference set-points, as follows:

Irefd =2

3

PrefVgd + QrefVgq

V 2gd

+ V 2gq

(5)

Irefq =2

3

PrefVgq + QrefVgd

V 2gd

+ V 2gq

(6)

These current references are then passed through the currentclosed-loop control to obtain the final voltage dq-axes refer-ences; feed-forward terms should be used to decouple the twoaxes, and should be considered for the difference between thevoltages after and before the ac filter. Neglecting Rd, a singleline diagram can be used to derive such a relation, as shownin Fig. 4, resulting in the following equations:

4

Lf Rf

CfVo Vg

agI

Fig. 4. Single line diagram of ac filter.

drefI

qrefI

dgI

qgI

p iK K s

p iK K s

fL

fL

+-

+

-

+-

+

+

drefV

qrefV

fR

fR

+

+

2

f fL C

f fR C

f fR C

2

f fL C

dgV

-

qgV

-

dgV

qgV

+

-

+

+

Fig. 5. Current closed-loop control.

Vod = Vgd(1−ω2LfCf )+IgdRf−IgqωLf−VgqωRfCf (7)

Voq = Vgq (1−ω2LfCf )+IgdRf +IgdωLf +VgdωRfCf (8)

The final current closed-loop control block shown in Fig. 5can be obtained from these equations.

The output of the current controller are the voltage refer-ences Vrefd and Vrefq , which are transformed back to theabc-reference frame to obtain the sinusoidal control signalsfor PWM scheme of the converter. More information onalternative converter controls such as grid-supporting and grid-following can be found in [15].

D. AC Filter

The ac filter should be designed properly to limit outputcurrent ripple and achieve an acceptable damping rate. Theac filter inductor Lf in a dc-ac converter can be determinedbased on the maximum acceptable current ripple, the dc linkvoltage, and the switching frequency as follows [16]:

∆Imax =Vdc/3

Lf4fs(9)

The ac filter capacitor Cf reactive power injection hasto be less than 5% of the converter rated power [16]. Aseries damping resistance is considered to prevent harmonicoscillations, and is assumed to consume 0.2% of the ratedpower as per:

Rd =V 2I

0.002PI−

√(V 2I

0.002PI

)− 1

(ωCf )2(10)

Also, a resistance Rf is added in series with the filterinductance to represent its parasitic resistance losses.

Voc

Vga

Vgb

Vgc

Lf Rfa

b

c

Inverter AC Filter Grid

Voa

Vob

Cf

Rd

Vdc

PLL

abcdq0

Active and

Reactive Power

Calculation

Current

closed-loop

control

Current Reference

Generator

Ig_dq

Q

P

Qref

abcdq0

Vref_dq

Vref_abc

Modulation Technique

Pref

V-F Reference

Generator

1

2Vref_dq

ref

2 1

refPLL

fref Vref

Vg_dq

PLL

Bat

tery

Ener

gy S

tora

ge

Syst

em S

chem

eM

easu

rem

ent

and C

ontr

ol

Sch

eme

agI

bgI

cgI

cgVbgV

agV

dqgVdqgI

abcrefV

dqrefV

dqrefV

aO

V

bO

V

cO

V

Fig. 6. Ideal dc link model of the BESS.

III. ESS APPROXIMATE MODELS

In the previous section, the different BESS components andtheir corresponding controls are discussed in detail. Mdelshave been proposed in the literature, in which some compo-nents of the detailed BESS model are eliminated to achievefaster simulation speed and/or reduce control complexity, asexplained next.

A. Ideal DC Link ModelThe schematic of this model is shown in Fig. 6, where the

battery model along with the buck-boost converter is replacedby an ideal dc source, thus fixing the dc link voltage. Hence,the controls related to the dc link are eliminated. The rest ofthe system remains the same, i.e. the inverter and the ac filterand the corresponding control are the same as explained inSection II.

B. Average ModelThe schematic of this model is shown Fig. 7, where the

entire switching system is replaced by ideal voltage sources,i.e. the battery, buck-boost converter, and dc-ac inverter areeliminated; this increases the simulation speed significantly.The control system for the battery remains the same as inSection II; the only change is that abc-reference signals aredirectly fed to the ideal voltage sources, and the PWM schemecontrols are eliminated.

C. Dependent Source Model (DSM)To identify and isolate the impact of switches on the

performance of the system, a model is proposed here in whichthe switches are replaced by dependent voltage and currentsources, as shown in Fig. 8. The control system remainsthe same as in Fig. 1. Two models are developed based onthe value of the dependent voltage and current sources, asexplained next.

5

Voc

Vga

Vgb

Vgc

Lf Rfa

b

c

Ideal Voltage Sources AC Filter Grid

Voa

Vob

PLL

abcdq0

Active and

Reactive Power

Calculation

Current

closed-loop

control

Current Reference

Generator

Ig_dq

Q

P

Qref

abcdq0

Vref_dq

Pref

V-F Reference

Generator

1

2Vref_dq

ref

2 1

refPLL

fref Vref

Vg_dq

PLL

Vref_abc

Bat

tery

Ener

gy S

tora

ge

Syst

em

Sch

eme

Mea

sure

men

t an

d C

ontr

ol

Sch

eme

aO

V

bO

V

cO

V

agI

bgI

cgI

cgVbgV

agV

dqgVdqgI

abcrefV

dqrefV

dqrefV

Fig. 7. Average model of the BESS.

1) Switching DSM: To emulate the performance of themodel with switches, the currents through and voltages acrossthe switches are analyzed in each switching state. For thebuck/boost converter, when Switch 1 is on and Switch 2 isoff in Fig. 1, Vd1n = Vdc, and the current through Switch 1 isIchop. Hence, VB and IB in Fig. 8, can be defined as follows:

Vd1n = VB = DS1 Vdc (11)

IB = DS1 Ichop (12)

where DS1 is the duty cycle of Switch 1.For the dc-ac inverter, when Switch 3 is on and Switch 4

is off, Van = Vdc, and the current through Switch 3 is Iga .Thus, Va and Ia in Fig. 8 can be defined as follows:

Van = DS3 Vdc (13)

Ia = DS3 Iga (14)

where DS3 is the duty cycle of Switch 3. Similarly for theother two legs of the inverter. Note that the internal resistancesof switches are modelled by Rin.

2) Average DSM: To eliminate the impact of high-frequency switiching, an averaging technique based on thefundamental frequency component is used here. Thus, byassuming that no current goes through Cf in Fig. 1 and usingKVL, the following equations can be derived:

Van = RfIga + LfdIgadt

+ Vga + VnN

Vbn = RfIgb + LfdIgbdt

+ Vgb + VnN

Vcn = RfIgc + LfdIgcdt

+ Vgc + VnN

(15)

A ∆-Y transformer usually connects the the inverter to therest of the grid, thus:

Vga + Vgb + Vgc = 0

Iga + Igb + Igc = 0(16)

Based on (15) and (16) the following equation can be obtained:

VaN = Van + VnN = Van − Van + Vbn + Vcn

3(17)

Similarly for the voltage on the other phases. In addition, DS3

in (13) and (14) can replaced by its low-frequency contents,i.e., its local average, as follows [17]:

DS3avg=

1

2+

1

2ma (18)

where ma is the modulating signal used to control DS3,normalized to the peak value of the carrier signal. Thus, from(17) and (18), the dependent source in phase a can be definedas follows:

VaNavg=

(1

2+

1

2ma

)Vdc−

32 + 1

2 (ma + mb + mc)

3Vdc =

1

2maVdc

(19)

And similarly for the dependent voltage sources in the otherphases.

Finally, using KCL, the following equation can be obtained:

Idc = Ia + Ib + Ic (20)

Thus, based on (20), (18), and (16), the dependent currentsource in phase a can be defined as follows:

Iaavg=

1

2maIga (21)

And similarly for the current source in the other phases.

IV. RESULTS

To effectively compare the performance of the three dif-ferent modeling approaches, a test system based on theCIGRE benchmark for medium voltage distribution networkintroduced in [18] is implemented in PSCAD/EMTDC, asshown in Fig. 9. The test system has a 1.3 MVA diesel-basedsynchronous machine, a 1 MW BESS, and a 1 MW windturbine, with the latter being modeled using an average modelsimilar to Fig. 7, and operated as a constant power sourcefor the short duration of time it is connected to the grid,since the disconnection of the wind turbine is used here asa disturbance. The diesel-based synchronous machine and itsexciter and governor are tuned and validated according theactual measurements from a commercial grade synchronousmachine in [19]. The BESS design parameters are shownin Table I. The loads are modeled using a voltage-sensitiveexponential model with a 1.5 exponent, which is a reasonablevalue for typical isolated microgrids [20]. The load demandand unbalance levels are different for each test scenariodiscussed next.

6

Voc

Vga

Vgb

Vgc

Vdc

Ichop Idc

Lf Rf

RB

EB

Lchopf Rchopf

Buck/Boost DC Link Inverter Grid

Voa

Vob

Battery

Cf

Rd

aO

V

bO

V

cO

V

agI

bgI

cgI

cgVbgV

agV

AC Filter

+

-

+-

+-

+-

n

Ib

Ic

Va

Vc

VbRin

Rin

Rin

Rin

VB

IB

N

Ia

Fig. 8. Schematic of a battery energy storage system with switches being replaced by dependent sources.

TABLE IESS DESIGN PARAMETERS

ESS ParametersLf Rf Cf Rd Vrateddc

0.166 mH 4.2 mΩ 626.8 µF 84.7 mΩ 750 vCdclink RB Lchopf fs VRMSL−L

20 mF 0.2 Ω 3.3 mH 3 kHz 460 v

1

2

3

4

511

10

9

8

7

6

12

13

14

1.0 km

1.2 km

4.9 km

2.0 km

3.0 km

1.7 km

0.2 km

0.6 km 0.5 km

0.6 km

WindDies.ESS

Load

Bus

Fig. 9. Modified version of CIGRE benchmark microgrid.

A. Case A: Grid-forming BESS in Balanced Operation

In this case, the diesel and wind generators are not con-nected. The system load is 950 kW of active power and 100kVar of reactive power, balanced among the three phases; att = 0.5s, the load is increased by 100 kVar, and at t = 11.5s,the reactive power load is again increased by 100 kVar; notethat the active power load is near the BESS rated power.

Figure 10 illustrates the active power, reactive power, and

instantaneous voltage of phase a at the PCC for average, idealdc link, and detailed model. Note in Fig. 10(a), before t =11.5s, the battery performance is satisfactory for all the threemodeling techniques. Due to switching in the system, there aresome fluctuations in the active and reactive power in the idealdc link and detailed models of the BESS. The Total HarmonicDistortion (THD) of the instantaneous voltage is 0.25% foraverage model, 3.92% for ideal dc link model, and 3.63% forthe detailed model.

After t = 11.5s, the system remains stable for the averageand ideal dc link models, whereas it shows unsustained oscil-lations for the detailed model. This is due to the fact that as theapparent power demand increases, the charging/dischargingcurrent of the dc link capacitor also increases, and beyonda certain value, the dc link capacitor voltage ripple becomessignificant, leading to system instability, as illustrated inFig. 11. Note that the chosen dc link capacitor is relativelylarge for a 1 MW BESS, and the active power is within therange for which the system is designed. Also, observe thatthe voltage magnitudes at the PCC are close to their nominalvalues in the average and ideal dc link models, as comparedto the detailed model of the BESS.

Figure 12 shows the BESS active power, reactive power,phase a RMS voltage at PCC, and the dc link voltage for theAverage and Switching DSMs. Observe that the performanceof the Average DSM is exactly the same as the performanceof the average model shown in Fig. 10, and the performanceof the Switching DSM is exactly the same as the performanceof the detailed model. These models are used in Section V toisolate and evaluate the impact of high frequency switchingon the stability of the microgrid.

B. Case B: Grid-forming BESS in Unbalanced Operation

In this case, the diesel and wind generators are not con-nected. The system load is 950 kW of active power and 100kVar of reactive power, with the load on phase c being twicethe load on phase a and b.; at t = 0.5s, the load is increasedby 100 kVar.

Figure 13 presents the active power, reactive power, andRMS phase voltages at the PCC. As seen from Fig. 13, the

7

2 4 6 8 10 12 14 16

0.80.9

1P

(MW

)

Average

2 4 6 8 10 12 14 16

0.80.9

1

P(M

W)

Ideal DC Link

0 2 4 6 8 10 12 14 16

0.80.9

1

P(M

W)

t(s)

Detailed

(a)

2 4 6 8 10 12 14 16

0.1

0.2

0.3

Q(M

VA

R)

Average

2 4 6 8 10 12 14 16

0.1

0.2

0.3

Q(M

AV

R)

Ideal DC Link

0 2 4 6 8 10 12 14 16

0.1

0.2

0.3

Q(M

AV

R)

t(s)

Detailed

(b)

2 4 6 8 10 12 14 16

0.95

1

Volt

(pu)

Average

2 4 6 8 10 12 14 16

0.95

1

Volt

(pu)

Ideal DC Link

0 2 4 6 8 10 12 14 16

0.95

1

Volt

(pu)

t(s)

Detailed

(c)

Fig. 10. Case A: (a) Active power, (b) reactive power, and (c) phase a RMSvoltage at the PCC.

0 2 4 6 8 10 12 14 16400

600

800

1000

Volt(v)

t(s)

Fig. 11. DC link voltage for the detailed model for Case A.

2 4 6 8 10 12 14

0.8

0.9

1

P(M

W)

DSM Switch.

0 2 4 6 8 10 12 14 16

0.8

0.9

1

P(M

W)

t(s)

DSM Av.

(a)

2 4 6 8 10 12 14

0.1

0.2

0.3

Q(M

VA

R)

DSM Switch.

0 2 4 6 8 10 12 14 16

0.1

0.2

0.3

Q(M

VA

R)

t(s)

DSM Av.

(b)

2 4 6 8 10 12 14

0.95

1V

olt

(pu

)

DSM Switch.

0 2 4 6 8 10 12 14 16

0.95

1

Vo

lt(p

u)

t(s)

DSM Av.

(c)

600

800

1000

Vo

lt(p

u)

DSM Switch.

0 2 4 6 8 10 12 14 16400

600

800

1000

Vo

lt(p

u)

t(s)

DSM Av.

(d)

Fig. 12. DSM Case A: (a) Active power; (b) reactive power; (c) phase aRMS voltage at PCC, and (d) dc link voltage.

system remains stable after the disturbance for the averageand ideal dc link BESS models; however, the system showsunsustained oscillations for the detailed model, since similarto Case A, the dc link capacitor voltage ripples cause thesystem instability, showing a similar behaviour as in Fig. 11.Comparing Case A and Case B for the same loading condition,it is noted that the same disturbance makes the unbalancedsystem unstable, whereas the balanced system remains stable.

The performance of the Average DSM has found to beexactly the same as the performance of the average model, and

8

2 4 6 8 10

0.8

1P

(MW

)

Average

2 4 6 8 10

0.8

1

P(M

W)

Ideal DC Link

0 2 4 6 8 10

0.8

1

P(M

W)

t(s)

Detailed

(a)

2 4 6 8 10

-0.10

0.10.2

Q(M

VA

R)

Average

2 4 6 8 10

-0.10

0.10.2

Q(M

AV

R)

Ideal DC Link

0 2 4 6 8 10

-0.10

0.10.2

Q(M

AV

R)

t(s)

Detailed

(b)

2 4 6 8 10

0.9

1

1.1

Vo

lt(p

u)

2 4 6 8 10

0.9

1

1.1

Vo

lt(p

u)

2 4 6 8 10

0.9

1

1.1

Vo

lt(p

u)

t(s)

VaAverage

VbAverage

VcAverage

VaIdeal DC Link

VbIdeal DC Link

VcIdeal DC Link

VaDetailed

VbDetailed

VcDetailed

(c)

Fig. 13. Case B: (a) Active power, (b) reactive power, and (c) RMS phasevoltages at the PCC.

the same for the performance of the Switching DSM comparedto the detailed model. Hence, these plots are not shown here.

C. Case C: Grid-forming BESS with Wind Generator Outage

In this case, the diesel generator is not connected, and thewind turbine is generating a constant 200 kW of active powerand 200 kVar of reactive power; at t = 1s, the wind generatoris disconnected. The system active power load is 1 MW andreactive power load is 350 kVar, with the load on phase cbeing twice the load on phase a and b.

1 2 3 4 5

0.8

0.9

1

P(M

W)

1 2 3 4 5

0.8

0.9

1

P(M

W)

0 1 2 3 4 5 6

0.8

0.9

1

t(s)

P(M

W)

Average

Ideal DC Link

Detailed

(a)

1 2 3 4 5

0.2

0.3

0.4

Q(M

VA

R)

1 2 3 4 5

0.2

0.3

0.4

Q(M

VA

R)

0 1 2 3 4 5 6

0.2

0.3

0.4

t(s)

Q(M

VA

R)

Ideal DC Link

Average

Detailed

(b)

1 2 3 4 5

0.80.85

0.90.95

11.05

1.11.15

Volt

(pu)

1 2 3 4 5

0.80.85

0.90.95

11.05

1.11.15

Volt

(pu)

0 1 2 3 4 5 6

0.80.85

0.90.95

11.05

1.11.15

t(s)

Volt

(pu)

VaAverage VbAverage VcAverage

VaIdeal DC Link VbIdeal DC Link VcIdeal DC Link

VaDetailed VbDetialed VcDetailed

(c)

Fig. 14. Case C: (a) Active power, (b) reactive power, and (c) phase RMSvoltages at the PCC.

Figure 14 illustrates the active power, reactive power, andRMS phase voltages at the PCC. Note that the system remainsstable for the average and ideal dc link models, but showsunsustained oscillations for the detailed model. Similar to CaseB, the dc link capacitor voltage ripple becomes greater thanwhat the system is designed for, after the wind generator isdisconnected.

The performance of the Average DSM has found to beexactly the same as the performance of the average model, andthe same for the performance of the Switching DSM compared

9

to the detailed model. Hence, these plots are not shown here.

D. Case D: Grid-supporting BESS With Wind Generator Out-age

In this case, the diesel generator is connected and is themaster voltage and frequency controller, with the governor andexcitation controls, and the BESS is providing 1 MW of activepower and half of the system reactive power demand. The windturbine is generating 500 kW of active power and 500 kVar ofreactive power, and the load active power is 2.1 MW and thereactive power is 1 MVar, balanced among the three phases.At t = 1s, the wind generator is tripped, and at t = 7s, theload reactive power is increased by 500 kVar.

Figure 15 shows the BESS active power, reactive power,and RMS voltages at the PCC. The diesel generator activeand reactive power, and the system frequency are illustratedin Fig. 17. As seen from these figures, the system is ableto retain its stability for the average and ideal dc link BESSmodels after the wind generator is disconnected, i.e betweent = 1s to t = 7s; however, the system loses its stability for thedetailed BESS model. In fact, since the battery has to increaseits reactive power generation after the disturbance, it passesa point where the dc link voltage cannot be maintained anylonger; as a result, the battery active power generation alsoshows large unsustained oscillations. Consequently, the dieselgenerator is not able to meet the demand level and the systemfrequency, and voltage collapses. The battery dc link voltagefor the detailed BESS model is shown in Fig. 16.

After the load reactive power is increased at t = 7s, thesystem continues to remain stable for the average modelof the BESS, but the system loses stability for the idealdc link model. This is an important observation, since theonly difference between these two models is the presence ofswitches.

The response of the Average DSM has found to be exactlythe same as the performance of the average model. TheBESS and diesel engine response of the Switching DSM isshown in Fig. 18 and Fig. 19. Observe that similarly to thedetailed model, the Switching DSM becomes unstable after thedisturbance at t = 1 s. However, its response is not exactlythe same as the detailed model, due to the different modelingapproaches; nevertheless, in both cases the system is instable,and thus they both lead to the same conclusions.

V. ANALYSIS

In the previous section, it was shown that when the systemwas pushed to its loading limits, different BESS modelsdemonstrated significantly different performances. These dif-ferences can be due to three main factors: the dc-link voltagedynamic, the high frequency switching, and the internal re-sistance of the switches. To effectively isolate the impact ofeach factor, first, an eigenvalue study is performed based ona signal-processing technique using the dc-link voltage signal.Second, the proposed DSM is used to eliminate the higherfrequencies and investigate the performance of the system.Additionally, the proposed models are compared in terms ofthe simulation computation times.

1 2 12 4 5 6 7 8 9 10 11 12 13 14

0.20.40.60.8

1

P(M

W)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

0.20.40.60.8

1

P(M

W)

0 1 2 3 4 5 6 7

0.20.40.60.8

1

t(s)

P(M

W)

Average

Ideal DC Link

Detailed

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14-0.5

0

0.5

1

Q(M

VA

R)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15-0.5

0

0.5

1

Q(M

VA

R)

1 2 3 4 5 6 7-0.5

0

0.5

1

t(s)

Q(M

VA

R)

Average

Ideal DC Link

Detailed

(b)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0.60.70.80.9

1

Vo

lt(p

u)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150.60.70.80.9

1

Vol

t(pu)

1 1 2 3 4 5 6 7

0.60.70.80.9

1

t(s)

Vo

lt(p

u)

Average

Ideal DC Link

Detailed

(c)

Fig. 15. Case D: BESS (a) Active power, (b) reactive power, and (c) RMSphase voltages at the PCC.

A. Small-Signal Analysis

Typically, in conventional power systems, eigenvalue anal-ysis is carried out by linearization of the power system modelaround an equilibrium point [21]. Such an approach requiresconsiderable simplifications limited to balanced systems, orthe development of detailed linearized models of the system[22], which may not be feasible in practical systems. In this

10

0 1 2 3 4 5 6 70

500

1000

1500

t(s)

Volt(v)

Fig. 16. DC link voltage for the detailed model - Case D.

paper, a modal estimation approach is utilized to identify thedominant eigenvalue of the system [23], [24]. In particular,the Steiglitz-McBride technique is utilized to identify thedominant eigenvalue of the system [25], [26].

To investigate the impact of dc-link voltage dynamics, thedc-link voltage signal of the detailed model in Case A isused to identify the critical eigenvalues before and after thesecond disturbance, using the MATLAB built-in Steiglitz-McBride function [27]. The critical eigenvalues of the systemare shown in Fig. 20, where it can be observed that thedominant eigenvalues are pushed to the right half plane afterthe second disturbance, resulting in the undamped oscillationsin Fig. 11.

B. Dynamic Analysis

Without loss of generality, the performance of SwitchingDSM and Average DSM are compared for Case A only inFig. 12. Note that the behaviour of the Switching DSM isexactly the same as the performance of the detailed modelshown in Fig. 10; thus, by comparing the performance ofthe Switching and Average DSMs, it is possible to isolatethe impact of high-frequency switching due to the switches,since the physical components of both DSMs are all the same,since only the values of the coefficients of the dependentsources in (13), (14), (19), and (21) are different, reflectingthe different switching content in the two models. As seen inFig. 12, the performance of the Average DSM is the same asthe performance of the average model, showing that includingthe dc link dynamics is not necessary in average modelingapproaches; furthermore, it confirms that is not possible tocapture the behaviour of the detailed model by neglecting theswitches and/or the impact of high-frequency swithcing. Inaddition, it can be concluded that including the dc link voltagedynamics is necessary but not sufficient to capture the accuratedynamic performance of the system.

With the Switching DSM, it is also possible to isolate andinvestigate the impact of Rin on the system’s performance.Thus, it was found that eliminating Rin has no considerableimpact on the performance of Switching DSM. Hence, itcan be concluded that the internal resistance of the switchesdoes not significantly contribute to the differences in theperformance of the different models.

C. Computation Times

The simulation times in Case A are compared for theproposed BESS models in Table II. The simulations were

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0.20.40.60.8

11.2

P(M

W)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

0.20.40.60.8

11.2

P(M

W)

0 1 2 3 4 5 6 7

0.20.40.60.8

11.2

t(s)

P(M

W)

Average

Ideal DC Link

Detailed

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0.40.81.21.6

Q(M

VA

R)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

0.40.81.21.6

Q(M

VA

R)

0 1 2 3 4 5 6 7

0.40.81.21.6

t(s)

Q(M

VA

R)

Average

Ideal DC Link

Detailed

(b)

1 2 3 4 5 6 7 8 9 10 11 12 13 14

5758596061

f(H

z)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

30405060

f(H

z)

0 1 2 3 4 5 6 7

30405060

t(s)

f(H

z)

Average

Ideal DC Link

Detailed

(c)

Fig. 17. Case D: Diesel generator (a) active power; (b) reactive power; and(c) system frequency.

performed in PSCAD/EMTDC 4.6.1.0 on a PC with anIntel(R) Xeon(R) CPU consisting of four 1.86 GHz processors,and 64 GB RAM. The computation times were obtained byaveraging waiting times for multiple 1 s simulation times,under similar server conditions, for the different models in thesame scenario. Note that the Switching DSM is faster than thedetailed model, while capturing its behaviour. Considering that

11

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

0.20.40.60.811.2

t(s)

P(MW)

(a)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15-0.5

0

0.5

1

1.5

t(s)

Q(MVAR)

(b)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150.50.60.70.80.911.1

t(s)

Volt(pu)

(c)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

500

1000

1500

t(s)

Volt(v)

(d)

Fig. 18. Switching DSM Case D: BESS (a) active power, (b) reactive power,(c) RMS phase voltages at the PCC, and (d) dc-link voltage.

TABLE IIAVERAGE COMPUTATION TIMES OF 1 S OF SIMULATION TIME IN CASE A

Detailed Ideal dc Average Switch. DSM Av. DSM50.6 s 43.4 s 24.8 s 39.4 s 27.0 s

the difference in computation times increases exponentiallyas the number of inverters in the system increases, those canbecome more considerable for larger systems. Finally, note thatthe Average DSM is slightly slower than the average model,yielding essentially the same performance.

VI. CONCLUSIONS

In this paper, first, three different BESS models werepresented and discussed: an average model that replaced theswitches with ac voltage sources, an ideal dc link model inwhich the battery and buck/boost converter were replaced byan ideal dc voltage source, and a detailed model that includedall necessary modeling and control details. In addition, toisolate and investigate the impact of high-frequency harmonicscaused by switchings, a new model was proposed in whichthe switches were replaced by dependent voltage and cur-rent sources, to capture the average and the fast switching

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1500.20.40.60.811.21.4

t(s)

P(MW)

(a)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 150

0.4

0.8

1.2

1.6

t(s)

Q(MVAR)

(b)

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1520

30

40

50

60

70

t(s)f(Hz)

(c)

Fig. 19. Switching DSM Case D: diesel generator (a) active power; (b)reactive power; and (c) system frequency.

-0.1 0 1.1320

120

Real

Imag(Hz)

Fig. 20. Dominant eigenvalue before and after instability in Case A.

behaviour of other models at adequate computational perfor-mance.

The following can be concluded from the presented studies:• During normal operating conditions, little differences

were observed between different models performances;the THD were higher for the fast-switching and ideal dclink models compared to the average models, due to theswitching in the system

• When the system loading was pushed to its limits, thefast-switching models lost their stability, while the ideal

12

dc link model and the average models remained stable.Also, it was demonstrated that the ideal dc link modellost its stability for certain loading levels for which theaverage models remained stable.

• Through eigenvalue studies, it was observed that dc linkdynamics contribute to the instability problem observedin the fast-switching modelings, due to the increase incharging/discharging current of the dc link capacitor.

• By comparing the performance of the DSMs, it can beconcluded that including the dc link voltage dynamicsis a necessary but not sufficient condition to capture theaccurate performance of the system. It was revealed thathigh-frequency dynamics also contribute the instabilityphenomena observed in the test scenarios.

• It was shown that the internal resistance of the switchesis not a major factor in the dynamic performance of thesystem.

• Comparing the computation time of the models, it wasnoted that during normal operating conditions, it may notbe necessary to include the dc link dynamics in averagingmodels. Thus, it can be concluded that when the BESS isin normal operating conditions, the average model is thebest option, since it significantly saves simulation timespeed, exhibiting adequate performance. However, if theBESS operates in grid-forming/grid-supporting modes,and the system is heavily loaded or unbalanced, usingaverage or ideal dc link models, or any other model thatneglects the dc link voltage dynamics and/or the highfrequency switching, do not properly capture the BESSbehaviour during or after a disturbance.

REFERENCES

[1] R. H. Lasseter et al., “CERTS microgrid laboratory test bed,” IEEETrans. Power Del., vol. 26, pp. 325–332, Jan. 2011.

[2] IEEE Guide for Design, Operation, and Integration of DistributedResource Island Systems with Electric Power Systems, IEEE Std. 1547.4,2011.

[3] J. M. Guerrero, F. Blaabjerg, T. Zhelev, K. Hemmes, E. Monmasson,S. Jemei, M. P. Comech, R. Granadino, and J. I. Frau, “Distributedgeneration: toward a new energy paradigm,” IEEE Ind. Electron. Mag.,pp. 52–64, March 2010.

[4] G. Delille, B. Francois, G. Malarange, and J.-L. Fraisse, “Energy storagesystems in distribution grids: New assets to upgrade distribution networkabilities,” in Proc. 20th International Conf. and Exhibition on ElectricityDistribution (CIRED) - Part I, Prague, Czech Republic, June 2009.

[5] G. Delille, B. Franois, and G. Malarange, “Dynamic frequency controlsupport by energy storage to reduce the impact of wind and solargeneration on isolated power system’s inertia,” IEEE Trans. Sustain.Energy, vol. 3, no. 4, pp. 931–939, Nov. 2012.

[6] A. Ortega and F. Milano, “Generalized model of VSC-based energystorage systems for transient stability analysis,” IEEE Trans. Power Syst.,vol. 21, no. 5, pp. 3369–3380, Sep. 2016.

[7] F. A. Inthamoussou, J. Pegueroles-Queralt, and F. D. Bianchi, “Controlof a supercapacitor energy storage system for microgrid applications,”IEEE Trans. Energy Convers., vol. 28, no. 3, pp. 690–697, Sep. 2013.

[8] J. Wu, J. Wen, H. Sun, and S. Cheng, “Feasibility study of segmentinglarge power system interconnections with AC link using energy storagetechnology,” IEEE Trans. Power Syst., vol. 27, no. 3, pp. 1245–1252,Aug. 2012.

[9] J. Fang, W. Yao, Z. Chen, J. Wen, and S. Cheng, “Design of anti-windupcompensator for energy storage-based damping controller to enhancepower system stability,” IEEE Trans. Power Syst., vol. 29, no. 3, pp.1175–1185, May 2014.

[10] Z. Jankovic, B. Novakovic, V. Bhavaraju, and A. Nasiri, “Averagemodeling of a three-phase inverter for integration in a microgrid,” inProc. IEEE Energy Conversion Congress and Exposition, Milwaukee,WI, Nov. 2014.

[11] J. W. Choi and S. K. Sul, “Inverter output voltage synthesis using noveldead time compensation,” IEEE Trans. Power Electron., vol. 11, no. 2,pp. 221–227, Mar. 1996.

[12] Y. Wang, G. Delille, X. Guillaud, F. Colas, and B. Francois, “Real-time simulation: The missing link in the design process of advancedgrid equipment,” in Proc. IEEE Power and Energy Soc. Gen. Meeting,Minneapolis, MN, July 2010.

[13] O. Tremblay, L.-A. Dessaint, and A. Dekkiche, “A generic battery modelfor the dynamic simulation of hybrid electric vehicles,” in Proc. IEEEVehicle Power and Propulsion Conference, Arlington, TX, Sep. 2007.

[14] S. Koenig, P. Muller, and T. Leibfried, “Design and comparison ofthree different possibilities to connect a vanadium-redox-flow-battery toa wind power plant,” in Proc. 13th CHLIE, Valencia, Spain, July 2013.

[15] S. M. Ashabani and Y. A. R. I. Mohamed, “New family of microgridcontrol and management strategies in smart distribution grids - analysis,comparison and testing,” IEEE Trans. Power Syst., vol. 29, no. 5, pp.2257–2269, Sep. 2014.

[16] M. Liserre, F. Blaabjerg, and S. Hansen, “Design and control of an lcl-filter-based three phase active rectifier,” IEEE Trans. Ind. Appl., vol. 41,no. 5, pp. 1281–1291, Sep. 2005.

[17] M. Restrepo, J. Morris, M. Kazerani, and C. Canizares, “Modelingand testing of a bidirectional smart charger for distribution system evintegration,” IEEE Trans. Smart Grid, vol. PP, no. 99, pp. 1–11, Mar.2016.

[18] K. Strunz, “Developing benchmark models for studying the integrationof distributed energy resources,” in Proc. IEEE Power Eng. Soc. Gen.Meeting, Montreal, QC, July 2006.

[19] M. Arriaga and C. A. Canizares, “Overview and analysis of datafor microgrid at kasabonika lake first nation (KLFN),” Hatch ProjectConfidential Report, University of Waterloo, Tech. Rep., Sep. 2015.

[20] G. Delille, L. Capely, D. Souque, and C. Ferrouillat, “Experimentalvalidation of a novel approach to stabilize power system frequency bytaking advantage of load voltage sensitivity,” in Proc. IEEE PowerTech,Eindhoven, June 2015.

[21] P. Kundur, Power System Stability and Control. New York, US:McGraw-hill Professional, 1994.

[22] R. H. Salim and R. A. Ramos, “A model-based approach for small-signal stability assessment of unbalanced power systems,” IEEE Trans.Power Syst., vol. 27, no. 4, pp. 1184–1190, Nov. 2012.

[23] R. H. Salim, R. A. Ramos, and N. G. Bretas, “Analysis of the small sig-nal dynamic performance of synchronous generators under unbalancedoperating conditions,” in Proc. IEEE Power Eng. Soc. Gen. Meeting,Minneapolis, MN, July 2010.

[24] E. Nasr-Azadani, C. A. Canizares, D. E. Olivares, and K. Bhattacharya,“Stability analysis of unbalanced distribution systems with synchronousmachine and DFIG based distributed generators,” IEEE Trans. SmartGrid, vol. 5, no. 5, pp. 2326–2338, Sep. 2014.

[25] K. Steiglitz and L. E. McBride, “A technique for the identification oflinear systems,” IEEE Trans. Automatic Control, vol. 10, no. 4, pp. 461–464, Oct. 1965.

[26] L. Ljung, System Identification: Theory for the User. Upper SaddleRiver, NJ: Prentice Hall, 1998.

[27] Matlab mathworks inc. [Online]. Available:https://www.mathworks.com/products/matlab.html

Mostafa Farrokhabadi (S’10) received his B.Sc. in Electrical Engineeringfrom Amirkabir University of Technology, Tehran, Iran, in 2010, his M.Sc.in Electric Power Engineering from KTH Royal Institute of Technology,Stockholm, Sweden, in 2012, and his Ph.D. in Electrical and Computerengineering from the University of Waterloo, ON, Canada, in 2017. He iscurrently a postdoctoral fellow at the Electrical and Computer EngineeringDepartment of the University of Waterloo. His research interests includesmodeling, control, and optimization in microgrids, mathematical modeling,and state estimation.

13

Sebastian Konig (M’17) received the Dipl.-Ing. and Dr.-Ing. degrees fromthe Karlsruhe Institute of Technology, Karlsruhe, Germany, in 2011 and 2017,respectively. His current research interest is electric energy storage and itsgrid integration. In his PhD thesis, he presents a model-based design andoptimization approach for the Vanadium Redox Flow Battery.

Claudio Canizares (S’85-M’91-SM’00-F’07) is a Full Professor and theHydro One Endowed Chair at the Electrical and Computer Engineering(E&CE) Department of the University of Waterloo, where he has held variousacademic and administrative positions since 1993. He received the ElectricalEngineer degree from the Escuela Politecnica Nacional (EPN) in Quito-Ecuador in 1984, where he held different teaching and administrative positionsbetween 1983 and 1993, and his MSc (1988) and PhD (1991) degrees inElectrical Engineering are from the University of Wisconsin-Madison. Hisresearch activities focus on the study of stability, modeling, simulation,control, optimization, and computational issues in large and small girds andenergy systems in the context of competitive energy markets and smart grids.In these areas, he has led or been an integral part of many grants and contractsfrom government agencies and companies, and has collaborated with industryand university researchers in Canada and abroad, supervising/co-supervisingmany research fellows and graduate students. He has authored/co-authored alarge number of journal and conference papers, as well as various technicalreports, book chapters, disclosures and patents, and has been invited to makemultiple keynote speeches, seminars, and presentations at many institutionsand conferences world-wide. He is an IEEE Fellow, as well as a Fellow of theRoyal Society of Canada, where he is currently the Director of the AppliedScience and Engineering Division of the Academy of Science, and a Fellowof the Canadian Academy of Engineering. He is also the recipient of the2017 IEEE Power & Energy Society (PES) Outstanding Power EngineeringEducator Award, the 2016 IEEE Canada Electric Power Medal, and of variousIEEE PES Technical Council and Committee awards and recognitions, holdingleadership positions in several IEEE-PES Technical Committees, WorkingGroups and Task Forces.

Kankar Bhattacharya (M95-SM01-F17) received the Ph.D. degree in elec-trical engineering from the Indian Institute of Technology, New Delhi, India,in 1993. He was in the faculty of Indira Gandhi Institute of DevelopmentResearch, Mumbai, India, from 1993 to 1998, and the Department of ElectricPower Engineering, Chalmers University of Technology, Gothenburg, Sweden,from 1998 to 2002. In 2003, he joined the Electrical and Computer Engineer-ing Department, University of Waterloo, Waterloo, ON, Canada, where he iscurrently a Full Professor. His current research interests include power systemeconomics and operational aspects.

Thomas Leibfried (M’96) received the Dipl.-Ing. and Dr.-Ing. degrees fromthe University of Stuttgart, Stuttgart, Germany, in 1990 and 1996, respectively.From 1996 to 2002, he was with Siemens AG, working in the powertransformer business in various technical and management positions. In 2002,he joined the Karlsruhe Institute of Technology, KIT as Head of the Instituteof Electric Energy Systems and High-Voltage Technology. He is the authorof various technical papers. Prof. Leibfried is a member of VDE and CIGRE.